Some distributed file systems heretofore use a network of controllers to manage transactions to and from a storage unit. Such controllers may replicate transaction logs amongst themselves. In the event a controller fails, the transaction logs may be used to determine the transactions that were in progress when the controller failed.